Postdoctoral Researcher (Post-Doc 4) at the Centre for Ius Commune, in the ERC Starting Grant p

Organisation/Company Adam Mickiewicz University, Poznań Department Centre for Ius Commune Research Field All Researcher Profile Recognised Researcher (R2) Positions Postdoc Positions Final date to receive applications 30 Sep 2026 - 23:00 (Europe/Warsaw) Country Poland Type of Contract Temporary Job Status Full-time Hours Per Week 40 Offer Starting Date 1 Dec 2026 Is the job funded through the EU Research Framework Programme?

Horizon Europe - ERC Reference Number  Is the Job related to staff position within a Research Infrastructure? No

Postdoc 4 in the Research Project PetrIUS:
Petrification of ius commune through printed paratexts
Full-time position, temporary contract with the first contract for 6 months, 36 months in total

The responsibilities of this position will include in general:

• Researching the printed legal paratexts.
• Development of digital tools for the study of paratexts using, among other methods, natural language processing techniques and deep learning.
• Contributing to the preparation of scholarly articles.
• Presenting research findings at conferences and academic seminars.
• Organizing seminars focused on the themes of the project.
• Cooperation within a research team (preference is given to on-site work; remote working arrangements are possible, provided that regular in-person attendance is maintained).

In particular, the Postdoc 4 will implement WP4 under the supervision of principal investigator.
WP4 Large-scale research of paratexts variants and their use in scholarly literature. This experimental part of the PetrIUS will use the possibilities offered by the natural language processing techniques to analyze the paratexts and their actual relevance through an automated large scale database survey. This type of research on incunabula is innovative and it requires that the specificity of the legal texts with multiple allegations and abbreviations be addressed.

The gain stemming out of the successful implementation of this WP outweighs these risks as it may constitute a contribution not only to our knowledge about the legal paratexts (their types, variants, and functions) but also to the methodological premises for further research on other marginalia in very different corpora of sources, natural processing of Latin, and using natural processing techniques for digitization of cultural heritage.

The implementation of this WP will rely on cutting edge scholarship and the expertise of Postdoc 4. In terms of methods, it is very likely that at the date of the implementation of this WP, there will be new methods available. In this case, we will follow the upto-date methodology. At this moment, the methods which seem particularly tailored for this WP are latent 4 semantic analysis (Landauer et al.

2007), latent Dirichlet allocation and topic modeling (Heibi et al. 2020), N-gram models (Arts et al. 2021), named entity recognition (Monroc et al. 2022), and large language models (Bamman and Burns 2020; Bonnell 2023). We might need to employ some additional digitizing and OCR-ing tasks (with tools provided by EMOP – (Use the "Apply for this Job" box below). ), because not all editions relevant to PetrIUS are digitized.

The implementation of the tasks will be preceded by the pilot implementations on the small subsets of data used for a quality check to guarantee the high efficiency of the outcomes.

There are three main tasks which are planned for this WP:
1. There will be trained algorithm(s) that will be able to distinguish between various types of paratexts based on their visual and spatial characteristics (Springmann and Lüdingex 2016; Reul, et al. 2017; Hill and Hengchen 2019). This task will require close cooperation between all team members to make sure that the algorithm will be efficient and effective. It will be necessary to provide some preliminary criteria inspired by the research on the printed paratexts and out of these intuitive guidelines to create the commands for the computational approach.

Ideally, the algorithm will be able to automatically indicate which parts of the OCR-ed page are paratexts with delimitation of their beginning and end. Surely, there will be some refinement needed,…